I'm trying to design an algorithm in NASM which is supposed to compute the nth Fibonacci number. I wrote some code, but, when run, it outputs only 1 and I cannot understand why. My idea was to call Fibonacci for n-1 and for n-2, where the parameter n is stored within AX register, and the result is stored at BX.
%include "io.inc"
section .text
global CMAIN
CMAIN:
mov ebp, esp; for correct debugging
xor eax, eax
xor ebx, ebx
push word 12
call fibonacci
pop word [trash]
PRINT_UDEC 1, ax
ret
fibonacci: push ebp
mov ebp, esp
mov ax, [ebp + 8];se incarca ax cu parametrul primit prin stiva
cmp ax, 0
jne l1
mov bx, 0
jmp l3
l1: cmp ax, 1
jne l2
mov bx, 1
jmp l3
l2: sub ax, 2;n-2
push ax
call fibonacci
pop dx;resotre the stack
mov [ebp - 2], bx
mov ax, [ebp + 8];se incarca bx cu parametrul prinit prin stiva
sub ax, 1;n-1
push ax
call fibonacci
pop dx;restore the stack
add bx, [ebp - 2]
l3: mov esp, ebp
pop ebp
ret
section .data
trash dw 1
Related
%include "asm_io.inc"
segment .data
segment .bss
segment .text
global asm_main
asm_main:
enter 0,0
pusha
call read_int
push eax
call fak_rekursiv
add esp, 4
call print_int
call print_nl
popa
mov eax, 0
leave
ret
fak_rekursiv:
enter 4, 0
pusha
mov eax, [ebp + 8]
cmp eax, 0
je ergebnis_1
cmp eax, 1
je ergebnis_1
mov ebx, eax
dec ebx
mul ebx
push ebx
call fak_rekursiv
pop ebx
ergebnis:
mov [ebp - 4], eax
ergebnis_1:
mov [ebp - 4], dword 1
popa
mov eax, [ebp - 4]
leave
ret
I am learning to code on NASM and I was trying to understand recursion through using coding factorial but I got confused quickly.
How can I use Recursion in NASM to code factorial algorithm?
actually I have coded up factorial just for fun earlier. And now I have dug through my files to find that code :) here you go, but it is more like a pseudo-code which I only stepped through debugger.
factorial:
push eax
test eax, 07FFFFFFEh
jz .exit
dec eax
call factorial
mul dword [esp]
.exit:
pop edx
ret
put different values in eax and step through it in debugger.
Not too familiar with NASM, but here's a MASM solution which calls the factorial function recursively.
factorial.asm
.386
.model flat, stdcall
option casemap :none
includelib \masm32\lib\msvcrt.lib
printf PROTO C, :VARARG
.data
fmt db "%u! = %u", 13, 10, 0
.code
factorial proc
cmp eax, 0 ; Special-case to account for 0! = 1
je retone
cmp eax, 1
je retnow
push eax
dec eax
call factorial
pop edx
imul eax, edx ; eax = eax * edx
retnow:
ret
retone:
mov eax, 1
ret
factorial endp
main:
; Loop from 0 to 12 (max) factorial for 32-bit code.
mov ebx, 0
num:
mov eax, ebx
call factorial
invoke printf, OFFSET fmt, ebx, eax
inc ebx
cmp ebx,13
jne num
ret
end main
Displays all factorials from 0 to 12
I am trying to implement Fibonacci sequence in assembly by using recursion. This is my first time of trying to implement recursion in x86 Assembly.
The code compiles fine but it gives wrong outputs. The output for 1 is 1, output for 2 is 0, output for 3 is 1, output for 4 is 2, output for 5 is 3.
Only output it gives correct when you plug 5 in.
Is there something wrong with the algorithm?
.DATA
n1 DWORD ?
prompt1 BYTE "Please enter the first value", 0
prompt3 BYTE "No negative numbers!",0
string BYTE 40 DUP (?)
resultLbl BYTE "The Fib is: ", 0
fib BYTE 40 DUP (?), 0
.CODE
_MainProc PROC
input prompt1, string, 40
atod string
test eax, eax
js signed
mov n1, eax
jmp procName
signed:
output prompt3, string
jmp end1
procName:
mov eax, n1
push n1
call fib1
add esp,4
dtoa fib, eax
output resultLbl, fib
end1:
mov eax, 0
ret
_MainProc ENDP
Fib1 proc
PUSH EBP ; save previous frame pointer
MOV EBP, ESP ; set current frame pointer
MOV EAX, [EBP+8] ; get argument N
CMP EAX, 1 ; N<=1?
JA Recurse ; no, compute it recursively
MOV ECX, 1 ; yes, Fib(1)--> 1
JMP exit
Recurse:
DEC EAX ; = N-1
MOV EDX, EAX ; = N-1
PUSH EDX ; save N-1
PUSH EAX ; set argument = N-1
CALL Fib1 ; compute Fib(N-1) to ECX
POP EAX ; pop N-1
DEC EAX ; = N-2
PUSH ECX ; save Fib(N-1)
PUSH EAX ; set argument = N-2
CALL Fib1 ; compute Fib(N-2) to ECX
POP EAX ; = Fib(N-1)
ADD ECX, EAX ; = Fib(N-1)+FIB(N-2)
exit:
MOV ESP,EBP ; reset stack to value at function entry
POP EBP ; restore caller's frame pointer
RET
Fib1 endp
END
The following is my code. The block in hex2dec works successfully for converting a single hexadecimal number to decimal number. It would be really helpful if someone could point out where I was going wrong in the use of array. Thanks.
DATA SEGMENT
NUM DW 1234H,9H,15H
RES DB 3*10 DUP ('$','$','$')
SIZE DB 3
DATA ENDS
CODE SEGMENT
ASSUME DS:DATA, CS:CODE
START:
MOV AX, DATA
MOV DS,AX
MOV DI,0
LOOP3:
MOV AX,NUM[DI]
LEA SI,RES[DI]
CALL HEX2DEC
LEA DX,RES[DI]
MOV AH,9
INT 21H
INC DI
CMP DI,3
JL LOOP3
MOV AH,4CH ; end program
INT 21H
CODE ENDS
HEX2DEC PROC NEAR
MOV CX,0
MOV BX,10
LOOP1:
MOV DX,0
DIV BX
ADD DL,30H
PUSH DX
INC CX
CMP AX,9
JG LOOP1
ADD AL,30H
MOV [SI],AL
LOOP2:
POP AX
INC SI
MOV [SI],AL
LOOP LOOP2
RET
HEX2DEC ENDP
END START
MOV AX,NUM[DI]
LEA SI,RES[DI]
LEA DX,RES[DI]
You are treating DI as an array index like we use in any of the high level languages. In assembly programming we only use displacements aka offsets in the array.
In your program, since the NUM array is composed of words, you need to give the DI register successively the values 0, 2, and 4.
ADD DI, 2
CMP DI, 6
JB LOOP3
Also it would be best to not treat the RES as an array. Just consider it a buffer and always use it from the start.
RES DB 10 DUP (0)
...
LEA SI, RES
CALL HEX2DEC
LEA DX, RES
A better version of HEX2DEC avoids the ugly prefixed "0" on the single digit numbers:
HEX2DEC PROC NEAR
XOR CX, CX <--- Same as MOV CX,0
MOV BX,10
LOOP1:
XOR DX, DX <--- Same as MOV DX,0
DIV BX
ADD DL, 30H
PUSH DX
INC CX
TEST AX, AX
JNZ LOOP1
LOOP2:
POP AX
MOV [SI], AL
INC SI
LOOP LOOP2
MOV AL, "$" <--- Add this to use DOS function 09h
MOV [SI], AL
RET
HEX2DEC ENDP
I've wrote assembly program which should count the below recursive function :
f(n) = f(n-1) + 2*f(n-2) - f(n-3)
For n = 0 and n = 1 it returns 1, and for n = 2 it should return 0.
But for this values program always returns 0, it looks like the below conditions were never met.
if0:
mov eax,1
jmp end
if1:
mov eax,1
jmp end
if2:
mov eax,0
jmp end
For any other value (greater than 2) I recives an segmentation falut.
Below is the entire code:
.intel_syntax noprefix
.globl main
.text
main:
mov eax, 3
push eax
call f
push eax
push offset msg
call printf
add esp, 8
mov eax, 0
ret
f:
push ebp
mov ebp,esp
add ebp,12
mov ebx,[ebp]
cmp ebx, 0
jz if0
cmp ebx, 1
jz if1
cmp ebx, 2
jz if2
lea ecx, [ebx-1]
push ecx
call f
pop ecx
push eax
lea ecx,[2*ebx-2]
push ecx
call f
pop ecx
pop eax
add eax,ecx
push eax
lea ecx, [ebx-3]
push ecx
call f
pop ecx
pop eax
sub eax,ecx
jmp end
if0:
mov eax,1
jmp end
if1:
mov eax,1
jmp end
if2:
mov eax,0
jmp end
end:
pop ebx
pop ebp
ret
msg: .asciz "Output = %d\n"
I have no idea what I do wrong.
EDIT: So, I've experimented with ebp and I've changed
add ebp,8
to:
add ebp, 16.
And, now it works for base conditions. It seems to me that there is problem with stack overflow, but I don't see it where is it.
Your pop ebx has no corresponding push, messing up your stack. There could well be other errors, but that's the one that jumped out at me.
Allright folks, let's hope this is an easy one: I need to access an array to achieve double-buffering (I use mode 13h) in 16-bit TASM. BUT: No matter if I use "OFFSET", "BYTE PTR [Array]", "BYTE PTR Array", or whatever I have tried already, the program reads/writes to the incorrect memory block, which is partly behind the actual start of the array.
Heres my (for now not really optimised and very messy) code:
.MODEL MEDIUM
.STACK
.DATA
XPos DW 0
YPos DB 0
Color DB 0
BoxX1 DW 0
BoxY1 DB 0
BoxX2 DW 0
BoxY2 DB 0
VPage DB 64010 DUP(0) ;TODO: Size *might* be incorrect.
PageSeg DW 0
.CODE
SetVGA13 PROC
MOV AX, 0013h ;Screen mode 13.
INT 10h ;Set screen mode to AX.
MOV AX, 0A000h ;Screen segment.
MOV ES, AX ;You can't affect segment registers
RET
ENDP
;-------DrawPixel---------------
; WORD XPos = x
; WORD YPos = y
; BYTE Color = colour
;-------------------------------
DrawPixel PROC
XOR AH, AH
MOV AL, [YPos]
MOV DX, 320
MUL DX
ADD AX, [XPos]
MOV DI, AX
MOV AL, [Color]
MOV ES, [PageSeg]
;ADD ES, DI
MOV ES:[DI],AL
;MOV ES:[DI],AL
RET
ENDP
DrawBox PROC
MOV CL, [BoxY1]
YLoop:
MOV BL, CL
PUSH CX
MOV CX, [BoxX1]
XLoop:
MOV [XPos], CX
MOV [YPos], BL
MOV [Color],CL
CALL DrawPixel
INC CX
CMP CX, [BoxX2]
JNZ XLoop
POP CX
INC CL
CMP CL, [BoxY2]
JNZ YLoop
RET
ENDP
WaitFrame PROC
PUSH DX
; Port #03DA contains VGA status
MOV DX, 03DAh
IN AL, DX
WaitRetrace:
; Bit 3 will be on if we're in retrace
TEST AL, 08h
JNZ WaitRetrace
EndRefresh:
IN AL, DX
TEST AL, 08h
JZ EndRefresh
POP DX
RET
ENDP
RestoreVideo PROC
; Return to text mode
MOV AX, 03h
INT 10h
RET
ENDP
ClearScreen PROC
XOR CX, CX
;MOV ES, [PageSeg]
ClearLoop:
MOV DI, CX
;MOV ES, [PageSeg]
MOV BX, OFFSET VPage
ADD BX, CX
MOV AL, [BX];VPage[DI];ES:[DI]
MOV [Color],AL
MOV AX, 0A000h
MOV ES, AX
MOV AL, [Color]
MOV ES:[DI],AL
INC CX
CMP CX, 64000
JNZ ClearLoop
RET
ENDP
Main:
;INITIALISE
MOV BX, OFFSET VPage
MOV [PageSeg],BX
CALL SetVGA13
;CALL MakePalette
MOV [BoxX1],33
MOV [BoxY1],33
MOV [BoxX2],99
MOV [BoxY2],99
;LOOP
GameLoop:
;DRAW
;CALL DrawBox
CALL ClearScreen
;CALL WaitFrame
;INPUT
MOV DX, 60h
IN AL, DX
CMP AL, 75
JNZ NotLeft
SUB [BoxX1],1
SUB [BoxX2],1
NotLeft:
IN AL, DX
CMP AL, 77
JNZ NotRight
ADD [BoxX1],1
ADD [BoxX2],1
NotRight:
CMP AL, 1
JNZ GameLoop
;END PROGRAM
Error:
;CALL ClearScreen
CALL RestoreVideo
MOV AH, 4Ch
INT 21h
END Main
This code shows a rainbow coloured box that you can move around with the left and right arrow keys,
;INITIALISE
MOV BX, OFFSET VPage
MOV [PageSeg],BX
That is my sad attempt to getbthe pointer to my buffer, but doesnt return the correct one
Sorry that my question was not done, i realised that when i got out of bed immediately for some reason.
Although this is not how I would code things, I'll offer up a couple of suggestions that may get you closer to resolving your problems.
When your executable loads the DS and ES registers originally point to the DOS PSP for your program. In your case you need to at least point DS to your DATA segment. The DATA segment that is filled in at run time by the DOS EXE loader can be referenced in your code by prefixing the segment name with an # (AT) symbol. So you can replace this code:
;INITIALISE
MOV BX, OFFSET VPage
MOV [PageSeg],BX
With this:
;INITIALISE
MOV BX, #DATA ; Set up DS with our program's DATA segment
MOV DS, BX
MOV [PageSeg],BX ; VPage is in DATA segment, move segment to PageSeg
In your DrawPixel code you'll need to add the VPage offset to DI. Replace this code:
XOR AH, AH
MOV AL, [YPos]
MOV DX, 320
MUL DX
ADD AX, [XPos]
MOV DI, AX
MOV AL, [Color]
MOV ES, [PageSeg]
MOV ES:[DI],AL
with:
XOR AH, AH
MOV AL, [YPos]
MOV DX, 320
MUL DX
ADD AX, [XPos]
MOV DI, AX
MOV AL, [Color]
MOV ES, [PageSeg]
ADD DI, OFFSET VPage ; We need to add VPage's offset
; from beginning of PageSeg
MOV ES:[DI],AL
Video mode 13h is 320x200x256 colors. The amount of video ram that is needed would be 320*200*1 = 64000. This could be changed from:
VPage DB 64010 DUP(0) ;TODO: Size *might* be incorrect.
to:
VPage DB 64000 DUP(0)
With these changes the program seems to display a rainbow box and moves left and right on the screen with the arrow keys.
There may be other bugs but I don't know what you were trying to achieve. This code could be simplified greatly. My changes were the minimal ones to work with the way you coded your program and make it somewhat functional.